Trying to use `NewtonTrustRegion` with inequality constraints produces StackOverFlow error

[MichaelBarmann]

(Just a note that I am new to this issue reporting thing; I'm writing this at the suggestion of @odow:https://discourse.julialang.org/t/how-to-properly-specify-jacobian-hessian-functions-of-inequality-constraints-optim/53870/3.

While testing out a constrained optimization problem with inequality constraints using IPNewton(), I noticed that changing the method to NewtonTrustRegion() caused it to throw the following error:

StackOverflowError:

Stacktrace:
 [1] optimize(::TwiceDifferentiable{Float64,Array{Float64,1},Array{Float64,2},Array{Float64,1}}, ::TwiceDifferentiableConstraints{var"#con_c!#89",var"#con_jacobian!#90",var"#con_hess!#91",Float64}, ::Array{Float64,1}, ::NewtonTrustRegion{Float64}, ::Optim.Options{Float64,Nothing}; inplace::Bool, autodiff::Symbol) at C:\Users\Michael\.julia\packages\Optim\CK6Dn\src\multivariate\optimize\interface.jl:148
 [2] optimize(::TwiceDifferentiable{Float64,Array{Float64,1},Array{Float64,2},Array{Float64,1}}, ::TwiceDifferentiableConstraints{var"#con_c!#89",var"#con_jacobian!#90",var"#con_hess!#91",Float64}, ::Array{Float64,1}, ::NewtonTrustRegion{Float64}, ::Optim.Options{Float64,Nothing}) at C:\Users\Michael\.julia\packages\Optim\CK6Dn\src\multivariate\optimize\interface.jl:147
 ... (the last 2 lines are repeated 28980 more times)
 [57963] ModelFit(; ODE_model::Function, ODE_vars::Array{String,1}, vars_to_fit::Array{String,1}, paramIC_dict::Dict{Symbol,Any}, data_dict::Dict{Symbol,String}, f_calc_ICs::typeof(f_ICs), N0::Int64, fit_range::Array{String,1}, date_obs_last::String, forecast_until::String, date_format::String, norm::typeof(mean_abs_rel_error), norm_weights::Array{Int64,1}, norm_scale::Int64, integrator_options::Dict{Symbol,Any}, optimizer_options::Dict{Symbol,Any}) at .\In[60]:317
 [57964] top-level scope at In[61]:15

Removing the inequality constraints made it it work, so it seems to be that NewtonTrustRegion doesn't support inequality constraints (is this true?). If so, it would be helpful to have a better error message, e.g.NewtonTrustRegion() doesn't support inequality constraints.

(On a somewhat tangential note, the Optim.jl documentation isn't quite clear about which methods support inequality, box constraints, and which do not. I know it has some examples, but it seems incomplete...Is this forum the place to post suggestions for the documentation? Or is there some other forum dedicated to that?)

[pkofod]

Thanks, I'll look into it. This is a good place to discuss documentation. Just open an issue with suggestions! Thanks